Originally Posted by cptomes
I can put an arbitrary delay on any of the outputs of the dcx.
Interesting concept. I think I'm just going to have to try it.
A digital delay is fixed with respect to frequency, which is generally what you want when aligning speakers since they don't tend to change position with frequency. But wavelength varies with frequency, thus fixed delay means varying phase shift with frequency. That won't achieve the desired goal here. You would create the equivalent of two monopoles spaced by some distance, which might smooth room response or not for a given location, but would also have some wonky interactions happening inside the enclosure.
An all pass filter, usually implemented in analog form, is a "poor man's" version of the delay when digital processing isn't available. This phase shift is frequency dependent, but can work OK for sub alignment since you can align phase at the crossover point, above and below which the crossovers filter one speaker or another reducing the effect of destructive interference away from the xo point.
But this "good enough" filter is just what we need here, because it has another useful property in that the filter shelves at 180 and 0 degrees, just like a low or high pass frequency shelving filter is flat above and below certain frequencies. That allows the sub to be 180 out of phase above a certain frequency and stay there... and be 0 degrees out of phase below a certain frequency and stay there. Dipole above some frequency, monopole below some lower frequency, with a smooth transition between.
I don't remember if the behringer allows all pass filter topologies. Generally not needed when you have digital delay available. A quick search or look through the user manual should answer that question.
To LTD, what you are after is something like a shelving filter with fixed offset shelf. I'm not sure if that is possible in analog implementation, but I doubt it without getting into multitap stuff. I'm not an EE, always have to look this stuff up. Probably possible with digital filters, I suspect requiring FIR as it is akin to linear phase crossover topology.